AI Coding, Ocean Freight, and Star Trek

AI coding will do more than make software cheap

It’s fascinating to me that a large-enough drop in the cost of just one thing can dramatically reorganize the large-scale systems that govern our lives: economies and societies. It’s this phenomenon that links my three topics today.

Ocean shipping is a perfect example. Ship-borne freight used to be packed in crates and bags, to be loaded and unloaded by longshoremen and stevedores, a process that was both slow and expensive. With the invention of containerized shipping, the ocean freight cost of the $140 made-in-Asia running shoes that you bought in the U.S. was just 70 cents, 1/200^th of your purchase price.

The extreme efficiency of ocean freight made world-spanning supply chains possible, globalizing the world economy, which in turn caused massive shifts in affordability (stuff is much cheaper), and large-scale job shifting among nations.

It’s interesting that this innovation that changed the world was created by one person. During the Depression, a young truck driver in his early twenties, Malcolm McLean, became frustrated by the inefficiency of freight handling. It took him twenty years to build his company and prove his concept, and in 1956, McLean’s first container ship sailed with 58 containers. His company grew to become Sea-Land, now part of Maersk.

The cost reduction factor from using AI to assist software creation dwarfs the savings that containerization yielded for shipping. The resulting shift in the software engineering job market is easy to anticipate, and already underway. But there will surely also be deeper impacts on our society — after all, in our technological civilization, software runs everything. Let’s explore one of those impacts.

Engineering any system requires choosing a tradeoff between flexibility and cost. It’s cheaper up front to build a rigid system; but then, as circumstances evolve — and they always do — it’s more expensive to retrofit or replace the system later.

AI coding assistants don’t just make it cheaper and faster to write new code; they make it cheaper and faster to modify existing systems. By itself, this would push the design of new systems to be less flexible. However, AI coding assistants also make it cheaper and faster to design and build systems that have greater flexibility. So AI coding doesn’t just move the choice along the tradeoff curve; it pushes the whole curve down, expanding the envelope of viable solutions.

Our technological civilization is also built on physical stuff. Recent advances are driving the hardware tradeoff curve down too. Gen AI, 3D printing, and new generations of reprogrammable microchips are making it possible both to design more-flexible hardware systems and to modify them more cheaply and faster later. For me personally, the aha! moment came when the 3D printer in my home produced a cell phone stand tailored to my personal preferences — and I didn’t have to wait for it to show up on my front steps. That one was designed by hand, but my next one will be designed by AI.

Back to software for a moment, then on to Star Trek. The other day, for fun, I partnered with an AI to build a website, and together we repeatedly modded it. It’s known that AI coding works best when the human is experienced, and since I’ve built websites, taught website programming, and used to be a web publisher, this project went well.

What caught me by surprise was the feeling my weekend sitebuild gave me: a true partnership of the AI and myself, a hybrid intelligence. It made me think of the Chief Engineer of the Enterprise. You know the plot: a life-and-death crisis erupts, and Scotty, working in partnership with the ship’s computer, saves the day by modding the starship into a new configuration. In an hour! Knowing the practical impossibility of speedily reengineering a complex system, I used to have to willingly suspend my disbelief. But after building the website with xAI’s Grok — two experts working as partners — modding a starship on the fly suddenly seems plausible.

Until now, our technological progress has been throttled by the fact that complex systems, once invested in, have to pay back their investment before they can be significantly changed with a further round of investment. Gen AI won’t just make it fast to build systems; it will make them fast to evolve at a pace previously unimaginable.

As unimaginable as it would have been a century ago to send your shoes from the other side of the planet for mere pennies.

© 2025 Boaz Super. All rights reserved.

Thank you to transportgeography.org for information about McLean.